ISM330DLC Drift Problem

[IKalh.1]

Hi all, 

I was wondering if anyone here has worked with the ISM330DLC IMU and was able to get positional displacement from it? I am trying to do this for one of my projects, but have run into drift issue. Any help is most appreciated! Thanks!

Thank you,

Imran

[niccolò · Accepted solution]

Hi @IKalhoro​ ,

the drift is unavoidable, because the noise of the sensor, even if small, is not zero.

you can try to set a threshold that is lower than the noise to trigger a movement, or implement a moving average, but you would lost accuracy for small distances.

If this is what you were looking for, please, flag this post as Best Answer =)

Niccolò

[IKalh.1]

Hi Niccolò,

Thank you for your quick response! I was curious to know what the maximum drift should be for this sensor?

I have some sample video and simulation data (IMU Drift Data) if you have the time to review to see if I am getting significant drift than should be allowed. Perhaps I am making a calculation error somewhere. It is important to note that I have put this IMU sensor on flexible PCB.

For Session 1, there was quite a lot of drift in the x-axis. I think that was because of the board experiencing a lot of flexion. For session 2, I put the board on a rigid body, and noticed much smaller drift in the x-axis when performing the same movement. I am currently looking to see if I can minimize sensor noise before feeding the data through our displacement algorithm.

I would appreciate any further insights you may have, and get to know more about your experience in extracting distances from this sensor.

Thanks again!

Imran

[IKalh.1]

Just to add on,

I decided to put the device on a motorized platform so that any human movements are negated. Here is a link to the setup (video). I have uploaded the accelerometer and gyroscope data during period of movement for your reference. Does the output make sense to you? Just want to be sure the sensor is working properly.

Accelerometer Output:

Gyroscope Output:

 

Thank you,

Imran

[Federica Bossi]

Hi @IKalhoro​ ,

According to DS the typical value for gyro noise is 75 mdps, while the typical values of accelerometer noise is 1.8mg @FS=2g.

looking quickly at your video and your data, these seem consistent.

[IKalh.1]

Great, thank you Federica for supplying that information!

Best regards,

Imran